2026 ASEE Annual Conference & Exposition

With support from SUNY Innovative Instruction Technology Grants (IITG), a retrieval-augmented, AI-powered conversational teaching assistant, eduBot, was developed and integrated into courses at SUNY Empire State University and SUNY Farmingdale State College. EduBot was developed to address common challenges in engineering and technical education, including limited access to timely academic support, early-course confusion, instructional scalability, and the need for responsible AI integration. The system was designed using instructor-approved materials and embedded within the learning environment to provide students with grounded responses related to course content, schedules, policies, and foundational concepts.

This paper reports findings from a matched pre/post survey study of students who used eduBot in an engineering course at Farmingdale State College. Twenty students completed both the pre- and post-surveys and were included in the matched analysis. Survey items measured student expectations and post-use perceptions related to usefulness, confidence, engagement, timely support, assessment preparation, user-friendliness, and willingness to use eduBot in other courses. Because the pre- and post-surveys used slightly different response formats, responses were converted to a standardized 0–100 scale before analysis.

Overall perceptions of eduBot improved from pre-survey to post-survey. The largest statistically significant gains were observed for using eduBot to access syllabus and schedule information and for receiving timely responses to student questions. Post-survey results also showed generally positive perceptions, with most students reporting favorable attitudes toward eduBot, its usefulness for course logistics, and its role in supporting engagement. These findings suggest that retrieval-augmented AI teaching assistants may be especially valuable for reducing logistical uncertainty and increasing access to timely course support in engineering education. Results further show that the eduBot model—designed and managed using the ADDIE framework—contributes to measurable improvements in student self-efficacy in content-intensive courses.